The transition from cohabitation to marriage in Spain: differences and determinants in same-sex and different-sex couples

Same-sex households and non-heterosexual people’s living arrangements are an expanding area of research in family studies. This contribution focuses on the transition from cohabitation to marriage of same-sex couples in comparison to different-sex couples in Spain, i.e., one of the pioneer countries of same-sex marriage that was legalized in 2005. In this particular context, we investigate to what extent same-sex couples and different-sex couples present similar marriage risks. Employing Event History Analysis and using data from the 2018 Spanish Fertility Survey, we find that, overall, same-sex couples have a significantly lower hazard of transitioning from non-marital cohabitation to marriage than different-sex couples. When restricting the sample to currently co-residing couples, the difference in the hazard of marriage between the two couple types remains negative but becomes insignificant. We thus disprove that same-sex couples have a higher incentive for marriage. These results are discussed in light of the different composition of same-sex vs different-sex couples in terms of their socio-demographic profile and motivations for marriage.


Introduction
While family has always been a crucial field in sociology and demography for its historical role of producing and raising children (Seltzer, 2019), in contemporary societies families are increasingly taking a wide variety of forms and structures, and the concept of family itself has evolved (Furstenberg et al., 2020).Among others, two phenomena in particular attracted the attention of scholars: the significant increase of non-marital cohabitation in Western countries (Noack et al., 2013) and the simultaneous diffusion of same-sex marriage and registered partnerships among non-heterosexual individuals (Digoix, 2020a).
Since the end of the last century, scholars started observing a strong decline in different-sex marriage and a change in the meaning of marriage (Cherlin, 2004;Kalmijn, 2007), while non-marital different-sex cohabitation increased (Klüsener et al., 2013;Lesthaeghe, 2020;Noack et al., 2013).Accordingly, a branch of research devoted to studying the characteristics, determinants and meanings of different-sex non-marital cohabitations flourished in demography (Perelli-Harris et al., 2014;Rose-Greenland & Smock, 2013;Sassler & Lichter, 2020).
Despite the growing attention on both non-marital cohabitation and same-sex unions, existing literature has mainly focused on non-marital cohabitation among different-sex couples and on marriage/registered partnerships among same-sex couples, thereby disregarding the experiences of non-marital cohabitation for same-sex couples.Yet, the two phenomena are linked.
If the gradual diffusion of non-marital cohabitation and non-marital childbearing resulted in a de-institutionalization of marriage among different-sex couples (Cherlin, 2004), the legalization of same-sex marriage is seen as reinforcing the salience of marriage as an institution (Cherlin, 2020).Same-sex couples are indeed subjected to unique triggers and incentives to transition to marriage, from legal recognition to social inclusion.In countries that grant parenting rights to same-sex couples, marriage may be important on the practical side for providing access to these rights otherwise denied.On the contrary, in countries that do not grant parenting rights to same-sex couples, marriage and registered partnerships may be avoided for fear of facing obstacles to the registration of the couple's (future) offspring.Also, marriage could retain a strong negative or ambivalent value for same-sex partners who can endorse it as a principle of equality but do not necessarily desire it for themselves (Haas & Whitton, 2015;Hull, 2019) because they perceive it as a reproduction of a heteronormative lifecourse (Digoix, 2020a(Digoix, , 2020b)).
The transition from cohabitation to marriage among same-sex couples, hence, may unfold differently compared to the transition observed among different-sex couples, and the drivers behind the transition to marriage likely differ across the two couple types, also depending on the context.In the Spanish context that will be studied in this contribution, parenting rights are granted to same-sex couples, hence, we may expect samesex couples to have a higher risk of transition to marriage compared to different-sex couples since marriage grants them easier access to parenthood and parenting rights.Yet, independently of the country context, we may observe a lower risk of marriage among same-sex cohabiting couples, linked to the fact that they are less attached to the institution of marriage compared to cohabiting different-sex couples.Finally, we may expect a different likelihood of transitioning from cohabitation to marriage among samesex vs. different-sex couples linked to their different demographic composition: partners in same-sex couples tend to be more educated than partners in different-sex couples, a factor that predicts lower hazards of marriage, but same-sex couples also tend to have a higher share of foreigner partners, that instead predicts a higher risk of marriage.
In order to control for environmental conditions related to demographic, economic, and socio-cultural factors (Burstein, 2003), this study focuses on a single country, Spain, where non-marital cohabitations are widespread and acceptance of same-sex couples and the LGBTQIA community is high (Abou-Chadi & Finnigan, 2019;Dotti Sani & Quaranta, 2022).The study of same-sex couples, indeed, underlines several obstacles which are not exclusively related, in the quantitative paradigm, to the access to reliable data on same-sex households (Compton, 2013;Cortina & Festy, 2020;Fischer, 2016;Fischer, 2022), but also to cross-national differences and fragmented legal frameworks.Spain was the third country worldwide to introduce same-sex marriage in 2005, hence allowing for a relatively long observation window which is suited for studying the transition to marriage among same-sex couples (Cortina & Festy, 2020).
This contribution compares the hazards of transitioning from cohabitation to marriage between Spanish same-sex and different-sex couples.The article has two main objectives: (1) to expand previous literature on same-sex couples' living arrangements by bringing non-marital cohabitation into the picture; (2) to compare cohabiting samesex and different-sex couples to better untangle-at least theoretically-possible motivations and differences in their transition to marriage.Since these couples already enjoy the benefits of living together, indeed, comparing potential differences in transitions could give us some clues on how differences in the meaning of marriage and cohabitation work in incentivizing (or not) marriage among these two groups.

Theoretical framework
From the early 1970s, European and North American countries started to decriminalize homosexuality (Waaldijk, 2004) and to gradually grant same-sex couples the right to form a union, whether marital or not, depending on the country context, hence ultimately granting same-sex couples increasing social inclusion via legal recognition (Digoix, 2020a;Saez, 2011).In 1989, Denmark became the first country in the world to legally recognize same-sex unions with the institution of the "registered partnership", followed shortly by other Scandinavian countries (Andersson et al., 2006;Waaldijk, 2004), but it was only in 2001 that same-sex marriage was introduced firstly worldwide in the Netherlands as an institution equal to different-sex marriage.

Same-sex marriage and legal rights
The majority of European countries have to date introduced some kind of legal recognition to same-sex unions, in the form of marriage or civil union.Same-sex couples, however, are generally granted fewer marital rights compared to different-sex couples (Saez, 2011;Waaldijk, 2020).In some countries, same-sex couples are excluded from parenthood and adoption, and the non-biological parent has no parental rights on children (De Rose et al., 2023).Even in countries where they have received full recognition from a legal point of view, same-sex couples may face opposition to their identities and relationships (Dotti Sani & Quaranta, 2020;Kennedy & Dalla, 2020).In most cases, samesex marriage has become often more a "subcategory" of different-sex marriage than an actual gateway to equality (Saez, 2011), a circumstance that is frequently emphasized by the presence of different rules for same-sex marriage and/or restricted access to certain rights (Saez, 2011;Waaldijk, 2004Waaldijk, , 2020)).One of the first issues that researchers must clarify when they want to study same-sex families in a certain context, thus, are potential differences in the procedures and rights attached to same-sex and different-sex marriage within that specific context.
Introduced by Waaldijk (2004) as a tool for investigating comparatively same-sex marriage, the Levels of Legal Consequences (LLC) measure the extension of legal rights and obligations-both between the partners and between the partners and the state-deriving from a formal act of marriage or registration of registered partnership.LLC, indeed, reveal more about the couples' situation than their marital status (Waaldijk, 2004(Waaldijk, , 2020) ) and, besides peculiar cross-country differences, it gives insights on five dimensions: implied mutual responsibilities, benefits for one partner implying responsibility for the other, immigration rights, benefits recognizing the couple as a unit, and parental rights.While different-sex cohabiting couples get access to all of the five dimensions when signing a marital certificate, it is not uncommon that countries grant to same-sex couples only some of the five dimensions or extend their LLC only after some time.Particularly, same-sex couples are more likely to be considered equal to different-sex couples in terms of economic rights than in terms of "controversial" rights such as parenting rights (Waaldijk, 2004(Waaldijk, , 2020)).Accordingly, even in some countries that have introduced registered partnerships (e.g., Hungary, Italy, Greece), same-sex couples are granted at most some economic benefits-such as inheritance, tenancy continuation or compensation for wrongful death-, but no parenting rights.While not investigated in the LLC framework (Waaldijk, 2004(Waaldijk, , 2020)), Spain provided almost automatically both economic benefits and parenting rights to same-sex couples with the amendment of the Civil code in 2005 which has equivalized same-sex and different-sex married couples.

Non-marital cohabitation and marriage
Cohabitation and marriage are intrinsically related both in the meaning people attribute them at the individual level (Noack et al., 2013) and in the way researchers interpret their trends at the contextual level (Manting, 1996).Cohabitation may be viewed as a less committed arrangement compared to marriage, a testing ground, a stage in the marriage process or as an alternative to marriage (Hiekel et al., 2014;Manting, 1996;Perelli-Harris et al., 2014;Waaldijk, 2004).Hence, the decision to transition from a nonmarital cohabitation to marriage is influenced not only by the personal characteristics of both partners, by how these characteristics are related to relationship trajectories (Rose-Greenland & Smock, 2013) and by intermediate elements and/or circumstances (e.g., career plans, couple projects, reproductive intentions) that vary throughout the life course (Cortina & Festy, 2020), but also by the partners' level of commitment, by what marriage means for them and by the value they attach to the institution of marriage.
Compared to marriage, cohabitation holds advantages and disadvantages: on the one hand, non-marital cohabitation allows for testing the relationship (Perelli-Harris et al., 2014), reduces the costs of union dissolution (Vignoli et al., 2016) and could be financially advantageous (Miho & Thévenon, 2020), on the other hand, cohabiters (and their children) have fewer legal rights and responsibilities than spouses who marry or register their partnership (Miho & Thévenon, 2020;Noack et al., 2013;Waaldijk, 2004).As nonmarital cohabitation has become widespread throughout the industrialized world, the important question is no longer why people cohabit instead of getting married, but why cohabiting couples marry (Noack et al., 2013).Cohabiting partners have already gained the advantages of living together in terms of economies of scale (e.g., via pooling of resources and sharing housing costs), so it should be asked what additional benefits they would acquire by transitioning to marriage.In these regards, based on the discussion of LLC in the previous paragraph, it could be argued that this question is even more meaningful when applied to same-sex couples.In the following paragraph, we hence discuss the meaning of marriage vs. cohabitation and the incentives to transition to marriage for same-sex couples.

Differences in the drivers of marriage between cohabiting same-sex and different-sex couples
The meaning of cohabitation and marriage are expected to change across relationship types and social groups (Reczek et al., 2009) and the reasons why people choose to transition from cohabitation to marriage are strongly dependent also on the characteristics of the couple itself.Hence, while the literature identified three main groups of drivers and determinants triggering the transition from cohabitation to marriage-namely socio-economic and demographic composition of the couple, values and practical consequences (Ishizuka, 2018;Kalmijn, 2007;Noack et al., 2013;Sassler & Lichter, 2020)differences in these determinants and how they work could be hypothesized between different-and same-sex couples.
Firstly, same-sex couples tend to be more selected on certain demographic characteristics that are usually associated with both lower and higher hazards of getting married.Same-sex couples are usually tertiary educated, a trait that is often associated with higher earnings potential, progressive values and thus more flexible living arrangements such as cohabitation (Domínguez-Folgueras & Castro-Martín, 2013), but they present also a higher proportion of foreign partners, who could be expected to have a preference for marriage as a gateway to citizenship rights (Andersson & Noack, 2010;Andersson et al., 2006).
Second, addressing values and symbolisms as determinants, qualitative studies highlighted how marriage for same-sex couples may retain a strong symbolic importance in a context of secularization and a higher symbolic value compared to different-sex couples (Cherlin, 2020;Kalmijn, 2007) being either a way to achieve relationship legitimacy or an act of "romantic activism" to encourage others to come out (Haas & Whitton, 2015).It is also true, still, that several same-sex couples seem to disregard marriage because they perceive it as a heteronormative institution or since it would involve a formal and public coming out and potentially undesired visibility (Cortina & Festy, 2020;Digoix, 2020a).
Lastly, dealing with practical determinants, it is expected that the legal consequences of marriage may retain a stronger power for same-sex couples than for different-sex couples.Indeed, while reproductive rights are "naturally" granted to different-sex couples, they are accessible to same-sex couples only at certain conditions as discussed in previous paragraphs.Thus, in countries that grant (at least some) parental rights to samesex couples only when married-which are the majority of Western countries nowadays (Waaldijk, 2020)-it is indeed expected that access to reproductive techniques, children's recognition and citizenship rights are a strong incentive to marriage (Cortina, 2016;Patterson, 2013;Pichardo, 2011).

Same-sex cohabitation and marriage in Spain
As discussed in the introduction, Spain is a suitable context to address our research question because it witnessed, in just over two decades, a considerable decline in religious marriages, an increase in non-marital cohabitation and in childbearing within cohabitation, while social values also changed dramatically, e.g., in terms of attitudes towards non-heterosexual people (Abou-Chadi & Finnigan, 2019;Dominguez-Folgueras & Castro-Martin, 2013;Dotti Sani & Quaranta, 2022;García-Pereiro, 2019).Such an early introduction of same-sex marriage in 2005 stimulated positive attitudes regarding LGBTQIA people (Waaldijk, 2020), so that Spain is nowadays one of the European countries with the highest acceptance of same-sex couples and marriage (Abou-Chadi & Finnigan, 2019).
While it is reasonable to believe that Spanish same-sex couples have different incentives to marry compared to different-sex couples, as discussed in the previous paragraph, it is indeed difficult to have clear expectations on whether same-sex couples have a higher or lower risk of transition from cohabitation to marriage compared to differentsex couples, given that drivers of marriage play in different directions.
First and foremost, according to benefits-related motivations, we could expect a higher risk of transition to marriage for same-sex couples with childbearing intentions as parenthood opportunities are higher in Spain for married couples.The implications of marriage for parenting are indeed of paramount importance for same-sex couples in Spain because while access to assisted reproduction technologies is open and free to women regardless of their civil status, until very recently marriage was required for female same-sex couples in order to register a newborn as a common child, and still is for having access to adoption (Hull, 2019;Saez, 2011).Hence, same-sex partners who are or intend to become parents may have a higher risk of marriage compared to different-sex cohabiting couples.
Instead, according to the partners' characteristics, we could expect both a lower risk of marriage-as same-sex couples tend to be more educated (Cortina & Festy, 2020) and more educationally homogamous (Cortina, 2016) than different-sex couples in Spainand a higher risk of getting married-as same-sex couples are more likely to include a foreign partner compared to different-sex couples (Cortina, 2016).These compositional elements are important as they may reinforce, compensate, or moderate the effects of other drivers on the risk of marriage.While there is mixed evidence regarding the association between education and cohabitation in Spain (Dominguez-Folgueras & Castro-Martin, 2013), it tends to be limited to different-sex couples.However, since a previous study indicated that highly educated homogamous couples have a lower likelihood of getting married (Maenpaa & Jalovaara, 2013), the distinctive educational profile of Spanish same-sex couples may suggest a preference for cohabitation.Differently, the possibility of acquiring Spanish citizenship also to individuals whose countries of origin do not recognize same-sex families is expected to work as a strong incentive for mixed couples (Girona et al., 2017;Saez, 2011).
Finally, according to value-based motivations, we could expect again a lower risk of marriage as same-sex couples are not necessarily attached to the institution of marriage as a personal aspiration (Hull, 2019) or a sign of commitment (Haas & Whitton, 2015); instead, they may perceive it as a heteronormative institution (Digoix, 2020a(Digoix, , 2020b)).
These three elements that drive the risk of transition from cohabitation to marriagecouples' characteristics, benefits-based motivations, and value-based motivations-play in different directions for same-sex couples.They do not only compete but also potentially compensate for each other.Because of their conflicting nature, thus, we do not formulate hypotheses.
One final remark regards same-sex couples within-group differences and, particularly, the gender composition of same-sex couples.Previous studies showed that the propensity to marry and divorce, as well as demographic characteristics such as education and premarital childbearing, differ among male and female same-sex couples (Andersson et al., 2006;Cortina, 2016;Kolk & Andersson, 2020).Moreover, in Spain, female same-sex couples can access Artificial Reproduction Technologies (ART) regardless of their civil status 2 (Pichardo, 2011), while male same-sex couples must be married to get access to parenting rights (i.e., adoption).Accounting for these differences, it is reasonable to expect that Spanish female and male same-sex couples differ not only compared to different-sex couples but also from one another.In this contribution, however, given the small number of same-sex couples in the sample, it was not possible to distinguish between female and male same-sex couples in the multivariate regression analyses.

Data and methods
We use cross-sectional data from the 2018 Fertility Survey collected by the Spanish National Statistical Office, collecting retrospective information on fertility, family and employment histories.For the first time, in 2018, the survey also interviewed a small sample of men (N = 2619), in addition to a large sample of women (N = 14,556), both aged 18 to 55 years old, which allows us to consider both male and female same-sex couples.

Samples
Three different analytical samples have been used in the analyses.First, for studying the hazard of transitioning from cohabitation to marriage in the survival analyses, we rely on 2 Even though female same-sex couples could access to ART regardless of their civil status, only married couples are automatically registered as the mothers of the baby on their birth certificate.Social mothers in cohabiting couples have to adopt the child to be recognized as parent.
a sample of partnered individuals, whether married or not, who, at the survey date, were co-residing with their same-or different-sex partner.Taking advantage of partnership histories collected by the survey, the sample also includes respondents' previous cohabiting unions (N = 7092).The second analytical sample used in the descriptive analyses was restricted to current unions, i.e., respondents who were co-residing with a partner at the time of the survey (N = 5071).This choice was forced by the limited information provided on the demographic characteristics of former partners, i.e., just their age.Finally, multivariate analyses were run only on a sample of female respondents, thus including all co-residing same-sex or different-sex couples, in which the main respondent was a woman (N = 4219).It is not possible, instead, to analyze co-residing male same-sex couples separately, given the small number of respondents in this sample (N = 19).
For all three analytical samples we excluded individuals whose marriage was celebrated before July 2005, i.e., when same-sex marriage became legal in Spain, to further increase the comparability between same-sex and different-sex couples.Finally, because we are interested in studying the transition from non-marital cohabitation to marriage, we excluded 194 respondents whose marriage date corresponds to the date when they started cohabiting.

Dependent and independent variables
The dependent variable used in the regression analyses is the hazard of transitioning from cohabitation to marriage at time t.Couples' hazard is calculated based on self-reported retrospective information on the time-expressed in months and yearsin which couples started their cohabitation and the time of their (eventual) marriage.Since registered partnerships have access to similar duties and rights as married couples (Waaldijk, 2020), we decided to code registered partnerships as married couples even though they were originally included as a separate category.For them, we considered as "time of marriage" the date when they registered their union.While we do not consider union dissolution as a competing risk, we decided to include also previous cohabiting unions.Such past unions exit the sample at risk when they marry (regardless of whether they then divorced or not) or when the union dissolved.Accordingly, time at risk is calculated by subtracting the year and month of marriage or union dissolution from the year and month when the non-marital cohabitation started.Couples were thus observed until they married, or, among unmarried previous cohabitants, until they broke up, or, if they are still in a non-marital cohabitation, until the time of the survey.
The main independent variable is the type of couple according to the couple's sex composition, namely whether the couple is a different-sex, female same-sex or male samesex one.The variable was constructed starting from two survey questions measuring the sex of the respondent and the sex of their partner.We acknowledge that this deductive approach for identifying same-sex couples, while largely used in demographic research, is subject to errors (Compton, 2013;Digoix, 2020a).Despite this, Fischer (2022) argues that identifying same-sex couples with household grids could be appropriate for studying partnerships (Fischer, 2016) while it is less reliable, for instance, for estimating the share of the LGBTQIA community in the population.

Control variables
The control variables included in the analyses are measured, when possible, at the couple-level, i.e., they refer to the characteristics of both partners.Controls include the following time constant variables: the level of education of both partners (both highly educated-i.e., above ISCED 5-at least one highly educated, none highly educated), their country of birth (both born in Spain, one born in Spain, both born abroad) and citizenship (both Spanish, one foreign, both foreign), household income in ranges (8 categories, ranging from 0 "No income" to 8 "5000 thousand or more"), and area of living (rural area, town, city).Because Spanish citizenship may have been acquired via marriage, we also include an interaction term between being in a same-sex couple and country of birth.Finally, we controlled for respondent's age, ranging between 18 and 55 years old.
Because employment history and the presence of children are important predictors of marriage transitions, we further include the following time-varying control variables: whether the respondent has ever been unemployed (until the survey date for cohabiting individuals and before marriage for those who transitioned to marriage); the length of unemployment spells experienced by the respondents during the time at risk, measured in months; whether the respondent has (at least) one biological child ever born (the variable was created from the question "What age did you have when your first biological children was born") and whether the child(ren) was born before marriage.

Methods
To estimate differences in transitions to marriage between same-sex and different-sex cohabiting couples we apply Event History Analysis and, specifically, a two-state model where marriage is treated as a non-repeatable event.Relying on respondents' selfreported retrospective information, we reconstruct their relationship history from the month when they started a non-marital cohabitation with their current or former partner (t0), until the month in which they transitioned to marriage, until they broke up with their former partner(s) or until survey date for current partners if they continued cohabiting (t1).Kaplan-Meier survival curves are computed for testing whether the risk and timing of the transition to marriage differ between both former and currently coresiding female and male same-sex couples and different-sex couples, while Cox proportional hazard models are used for estimating the effect of covariates on the transition to marriage among female same-sex and different-sex currently co-residing couples.Due to the small number of currently co-residing male same-sex couples in the dataset (N = 19), indeed, multivariate analyses will be conducted only comparing female same-sex couples with different-sex couples where the respondent is female.
In the multivariate analyses, we test the assumptions and expectations discussed in the theoretical framework by looking at the main effect and by controlling for compositional elements by adding controls.Unfortunately, controls for motivations derived from values could not be operationalized in the analysis.

Results
Table 1 shows the distribution of marital status across all the couples considered in the first analytical sample, i.e., different-sex and same-sex couples who were ever in a nonmarital cohabiting union.Half of different-sex couples (52.57%) transitioned to marriage compared to less than one-third of female (30.91%) and male (30.23%) same-sex couples.When considering only co-residing couples (see Table 5 in the Appendix), the proportion of married same-sex couples is higher than the one observed if both former and currently co-residing couples are considered.This higher proportion of married samesex couples is a crucial element to consider when interpreting the results of the Cox regression analysis.
Figure 1 reports the Kaplan-Meier survival curves for the probability of continuing to cohabit vs. transition from cohabitation to marriage among all couples, both currently co-residing or couples that were in a non-marital cohabitation in the past.Female same-sex couples in particular display lower hazards of transition to marriage.Similar evidence has been found in the descriptive statistics: female same-sex couples display a lower incidence of marriage (0.005) and a longer median survival time in non-marital cohabitation (63 months) than different-sex couples (see Tables 6 and  7 in the Appendix).Differences are statistically significant according to the log-rank test (not shown).Evidence on male same-sex couples is less robust as the subsample counts only 43 couples, but it is interesting to notice how the curve goes down to zero.This drop could be explained by the fact that all the male same-sex couples who  remain cohabitant in the sample did not experience such a long time at risk (i.e., the longest time at risk is 216 months).Differences between male same-sex couples and different-sex couples seem less consistent, particularly due to the small number of couples in the male sample.Similarly to what was found in other countries (Andersson & Noack, 2010;Andersson et al., 2006), same-sex couples in our sample tend to be, on average, more educated than different-sex couples (i.e., 30.36% of female and 15.79% of male same-sex couples have neither partner with high education vs. 32.54% in different-sex couples), more diverse in terms of age homogamy (i.e., 50% of female and 56.25% of male same-sex couples have more than 3 years of age difference compared to the 42.72% of different-sex couples), more likely to live in cities (about 57% for both female and male same-sex couples compared to the 50% of different-sex couples), while they are, respectively, almost two times (67.86%) for female same-sex couples and three times (84.21%) for male same-sex couples more likely to be childless compared to different-sex couples (31.5%).Interestingly, comparing female same-sex and male same-sex couples, the migrant background of the latter seems to be more heterogeneous (only 57% of male same-sex couples are composed of two natives) and to have a higher household income compared to both female same-sex couples and different-sex couples.Full descriptive statistics are provided in Tables 2 and 3.
Next, Cox proportional hazard regression investigates the association between socio-demographic characteristics and the hazard of marriage for currently co-residing female same-sex and different-sex couples where the respondent is a woman.As a robustness check, we run the regression on all co-residing same-sex couples regardless of whether they are female or male: results, reported in Table 8 in the Appendix, are robust.Model 1, i.e., the null model in Table 4, shows that same-sex couples have a lower hazard of transitioning to marriage (− 0.24) compared to different-sex couples with a 95% confidence interval that ranges from 0.64 to 1.2.Even though the difference is not statistically significant, this evidence is in line with the descriptive results in Fig. 1.Similarly, the difference between female same-sex and different-sex couples remains not statistically significant in Model 2 in Table 4, i.e., the full model including all control variables.Therefore, we find no evidence that same-sex couples are more likely to marry than different-sex couples as a result of stronger motivations related to benefits (Table 4).
In Model 3, in order to control for couples' characteristics, we further include interactions between the main explanatory variable (same-sex vs. different-sex couple) and partners' education, country of birth and citizenship.No significant interaction was found: the higher risk of marriage expected because of heterogeneous migrant background, and the lower risk expected by higher education did not find support.
Other results, particularly those related to having a child and unemployment, are in line with expectations and previous literature (Baizán et al., 2003;Cortina & Festy, 2020;Manning & Smock, 1995).While the direction of the coefficient for the variable "having a child before marriage" is unexpected (negative, − 1.87***), still it could be argued that these children could be born in previous relationships (thus before the actual cohabitation), but we don't have enough information to test it.
The association with educational level is not significant and marginal.Mixed couples in which one partner is born abroad or couples with at least one foreign citizenship have a higher and significant risk of transition to marriage compared to natives and people with Spanish citizenship.The controls for experience of unemployment are negatively associated with the risk of marriage while older people display a lower risk of transitioning to marriage.The fact that other determinants of marriage are in line with previous literature contributes to sustaining the overall reliability of these data and thus provides some support to the consistency of the findings related to same-sex couples.

Study limitations
The main limitation of this study is the small sample of same-sex couples in the dataset that prevented unpacking observed differences between female and male samesex couples in the transition to marriage and testing for interaction effects between couple types and the presence of children (as same-sex couples with children are a minority in the sample).Among the restricted sample used in the regression analyses, in particular, the small number of co-residing female same-sex couples does not allow for reliable estimations on characteristics or compositional effects that could be associated with higher or lower hazards of marriage.Accordingly, estimates are very uncertain and could also imply that cohabiting same-sex couples do not substantially differ from different-sex couples in their marriage hazards.
Other limitations are related to how same-sex couples were identified, i.e., based on the respondent's self-reported information about their own sex and the sex of their partners.Errors in the household grid matrix, indeed, could lead to biases such as, respectively, an underrepresentation of same-sex couples and hence an underestimation of real effects, or the estimation of a fictitious effect if different-sex couples are coded as same-sex ones by mistake.

Conclusions
Family demography is increasingly interested in the LGBTQIA population (Andersson et al., 2006;Boertien & Vignoli, 2019;Cortina, 2016;Cortina & Festy, 2020;Digoix, 2020b;Evertsson et al., 2021;Evertsson & Malmquist, 2023;Kolk & Andersson, 2020;Lau, 2012;Monaco & Nothdurfter, 2022;Perales & Baxter, 2018;Ruiz-Vallejo & Boertien, 2022), but this interest is frequently challenged by a lack of suitable data, with few exceptions such as studies based on primary-collected data (e.g., Monaco & Nothdurfter, 2021;Lelleri et al. 2008), on population registers for the countries where these are available (Evertsson & Malmquist, 2023;Evertsson et al., 2021;Kolk & Andersson, 2020;), or on representative (Boertien & Vignoli, 2019;Cortina, 2016; Cortina & Festy, 2020) or non-representative (Day et al., 2018;Lelleri et al., 2008) secondary data, again, for the countries where these are available.Most current studies, particularly those focusing on same-sex couples, rely on small samples available via nationally representative surveys.Using the same strategy, this contribution investigates the transition from non-marital cohabitation to marriage among a (small) sample of same-sex couples in Spain using the nationally representative Fertility Survey.Despite the obvious limitations linked to the small sample of same-sex couples in our data, the results of this study contribute to the growing body of family demographic literature on same-sex couples.Our research adds to the literature on samesex couples living arrangements by investigating same-sex couples' transition to marriage among a demographic understudied group, i.e., cohabitating same-sex couples.Using retrospective information and Event History methodology, we analyze the association between hazards and couples' characteristics, while also advancing the theoretical discussion on same-sex couples' drivers to marriage and differences in the cohabitation-marriage trade-off.In particular, focusing on cohabitants and thus on those people who already share the benefits of living together, our results underline that no differences seem to occur with different-sex couples in the risk of marriage and, if anything, they are less likely to transition to marriage.
To our best knowledge, no prior studies have ever tried to investigate this topic with this methodology and the only comparable research have investigated either the crude marriage rate of same-sex couples (Kolk & Andersson, 2020) or the probability of being in a non-marital cohabitation or being married (Cortina, 2016;Cortina & Festy, 2020;Manning et al., 2016), thus without looking at the transition itself and/or restricting the analyses to cohabiting couples.While, as expected, the crude marriage rate of same-sex couples is decisively lower than that of different-sex couples, previous findings on cohabitation and marriage found that same-sex couples have indeed a higher risk of being in a non-marital cohabitation compared to different-sex couples (Cortina, 2016;Cortina & Festy, 2020).Cortina and Festy (2020) explained samesex couples' lower propensity to marry in relation to their lower likelihood of having children-a compositional factor that is controlled for in our analyses.Other studies, primarily qualitative ones, suggest that marriage is indeed regarded among same-sex couples as important for providing access to legal benefits, but its symbolic value is ambivalent (Haas & Whitton, 2015;Hull, 2019;Reczek et al., 2009) and cohabitation is already perceived as an index of high commitment and a way to "being a family" (Haas & Whitton, 2015;Rostosky et al., 2016).
By focusing on cohabiting couples-and so accounting for part of the variance between same-sex and different-sex couples-our results support the view that samesex couples may be more likely to remain cohabitant than transition to marriage.The direction of the association is indeed negative-indicating a lower risk of transition to marriage as suggested by literature-, but the non-significance of the finding could suggest that the competitive drivers to marriage among same-sex cohabiting couples at least partially compensate one for each other.Further research is needed to understand the consistency of our findings and, potentially, to better explore the drivers of marriage among cohabiting same-sex couples.
Still, it is worth mentioning that our results could not hold for male same-sex couples and may be peculiar to the Spanish context.The early introduction in 2005 of same-sex marriages in Spain, indeed, stimulated positive attitudes regarding LGBT-QIA people (Waaldijk, 2020), so that Spain is nowadays one of the European countries with the higher acceptance of same-sex couples and same-sex marriage (Abou-Chadi & Finnigan, 2019;Dotti Sani & Quaranta, 2022).Accordingly, in Spain, marriage among same-sex couples may not be perceived as a way to achieve higher social recognition-as it may be the case in other countries-and thus the preference for marriage could be similar among same-and different-sex couples in this context.

Table 1
Type of couples by marital status-all couples included both currently co-residing and past couples(INE, 2018)

Table 2
Descriptive statistics of the categorical control variables included in the analyses by couple types-only currently co-residing couples(INE, 2018)

Table 3
Descriptive statistics of the continuous control variables included in the analyses by couple types-only currently co-residing couples.(INE,2018)

Table 4
Cox regression on the proportional hazards of transitioning from non-marital cohabitation to marriage among cohabitants by couple type-only currently co-residing female same-sex couples and different-sex couples where the respondent is woman.(INE,2018)

Table 5
Type of couples by marital status-only current co-residing couples are considered.(INE,2018)

Table 6
Summary of the survival data by couple types-all couples included both currently co-residing and past couples.(INE,2018)

Table 7
Median time at risk and incidence rate by couple type-all couples included both currently co-residing and past couples(INE, 2018)

Table 8
Cox regression on the proportional hazards of getting married among cohabitants by couple type-only currently co-residing couples without distinguishing between female-and malesame-sex couples.(INE,2018)